Data Warehouses Around the World Help Create an AI Supercomputer
Cernescu Andrei / 2 years ago
There has been a lot of talk about artificial intelligence as of late, particularly in regards to the dangers that it could pose. However, despite warnings from the likes of Stephen Hawking and Bill Gates, numerous companies around the world are striving to create a perfect AI, including Google. However, Google could run into some serious competition on the AI market thanks to a private company called Sentient, which has no more than 70 employees.
The company revealed that it could assemble enormous computing systems to power artificial-intelligence software by using an ingenious and cost-effective method: tapping into the unused power of data warehouses around the world. Data warehouses are built to cope with enormous demands and surges of traffic that don’t occur on a daily basis. Consequently, some of their computers are idle, for the most part, which allows Sentient to use them for its own computing needs. The company pays a data-center operator for access to the spare machines, and in the end, it can gather up to one million processor cores acting in unison in order to solve the same problem. According to the main architect for distributed computing at Sentient, Adam Beberg, Google’s largest machine-learning systems cannot reach this enormous scale.
Founded in 2007, Sentient has recently been trying to adapt its own methods to work with a type of AI named “Deep Learning.” Deep Learning has enjoyed breakthroughs in speech and image recognition, which is why it has caught the eye of tech giants such as Facebook and Google.
Thank you Technologyreview for providing us with this information.